近日，上海人工智能实验室联合商汤科技、香港中文大学和复旦大学，正式发布了新一代大语言模型书生·浦语2.0。这一突破性的进展标志着人工智能在语言处理领域的一大步进，书生·浦语2.0的核心理念是回归语言建模的本质，通过提升语料质量和信息密度，实现了模型基座语言建模能力的显著提升。
书生·浦语2.0的一大亮点是其强大的上下文处理能力，能够一次性接收并处理约30万汉字的输入内容，支持高达200K token的上下文长度。这意味着该模型能够在长文本中准确提取关键信息，实现“大海捞针”式的信息检索，极大地提高了语言模型的实用性和效率。
此次发布的书生·浦语2.0，不仅展示了上海人工智能实验室在自然语言处理领域的深厚实力，也为未来的人工智能应用提供了新的可能性。随着技术的不断进步，我们有理由相信，书生·浦语2.0将在智能对话、机器翻译、文本摘要等多个领域发挥重要作用，推动人工智能技术的进一步发展。
英文标题：Shanghai AI Lab Launches ShuSheng·PuYu 2.0
英文关键词：Language Modeling, Large Model, Long Text Processing
英文新闻内容：
The Shanghai Artificial Intelligence Laboratory, in collaboration with SenseTime, the Chinese University of Hong Kong, and Fudan University, has officially released ShuSheng·PuYu 2.0, a new generation of large language model. This breakthrough signifies a significant step forward in the field of artificial intelligence, particularly in natural language processing. The core philosophy of ShuSheng·PuYu 2.0 is to return to the essence of language modeling, enhancing the quality of corpus and information density to substantially improve the model’s language modeling capabilities.
One of the highlights of ShuSheng·PuYu 2.0 is its powerful context processing ability, capable of receiving and processing input content of up to 300,000 Chinese characters at once, supporting a context length of up to 200K tokens. This enables the model to accurately extract key information from long texts, achieving a “needle in a haystack” capability in information retrieval, significantly enhancing the practicality and efficiency of language models.
The launch of ShuSheng·PuYu 2.0 showcases the profound strength of the Shanghai AI Lab in natural language processing and opens up new possibilities for future applications of artificial intelligence. As technology continues to advance, it is expected that ShuSheng·PuYu 2.0 will play a crucial role in intelligent dialogue systems, machine translation, text summarization, and other fields, propelling the development of artificial intelligence technology forward.
【来源】https://www.thepaper.cn/newsDetail_forward_26040295
Views: 1